devchallenge

☠️ A support agent that reads a poisoned Sanity help center and still doesn't get tricked into moving money. Sanity Knowledge Base + Claude + taintgate. #sanitychallenge

Four-layer governance harness for Hermes Agent: budget cap, egress allowlist, audit trace, structured output. Built for the DEV Hermes Agent Challenge.

Todo list CLI

Telegram bot that reads bills and IOUs with local Gemma (Ollama) and keeps reminding until you pay. Durable reminders on Temporal, agent traces in Sentry. Built for the DEV Hacktoberfest 2026 Weekend Challenge.

Deterministic, pure-Python agent that decides whether retrieved claims from Sanity actually conflict — built for the Sanity Challenge, Path One.

साथी · An offline Hindi video-call companion. Gemma 4 hears her, sees what she holds up to the camera, and talks back, all on her laptop.

SENTINEL.EXE: An AI-integrated Tower Defense game built for the 2026 Google Portfolio Challenge. Defend the Human History Core against viruses using dynamic lore and tactical advice powered by Gemini 3 Flash.

MemoryMate — An AI built to help a friend remember the little things.

Local Open-Source AI Coding Interviewer

Resolução do desafio criado pela @Lorenalgm da DevChallenge

Can an LLM tell a Kademlia eclipse attack from an honest DHT lookup? A Kaggle benchmark where the obvious clue (IP clustering) and the real clue (ID distribution) disagree.